Possible confounds and their resolution in multivariate genetic analyses: comment on Bussell et al. (1999).

Assumptions behind the twin, sibling, and stepsibling design used in the Nonshared Environment in Adolescent Development study are considered. The D. A. Bussell et al. (1999) analyses of behavior in relationships indicate effects of both genetic and common-environment factors. Possible real and artifactual sources of the common environment include assortative mating, sibling interaction, failure of the equal environments assumption, and contaminated measures. Another potential source, specific to this study, is the effect of an individual in the mutual ratings of relationship with sibling, which would lead to a larger common-environment effect across variables than within variables. This alternative model did not indicate that there was a mutual rating effect. Univariate analyses confirm that the common-environment effect is present for both sibling and parent-child relationships. Most other potential sources of common environment, such as sibling interaction, do not account for the data.